Transformation, insecurity, and uncontrolled automation: frames of military AI in the EU AI strategic discourse

Despite the dual-use nature of AI, the emerging EU AI policy has excluded the military from the initial policy scope, by introducing it as a single-market issue and a predominantly civilian matter. However, both the existing EU tools for supporting developments in military AI and the existing literature on dual-use technologies suggest that the military realm is actually present, despite claims to the contrary. This tension raises a question: is the EU manoeuvring between the civil and military elements of AI in order to sideline the conversation on the military role in its self-identification and approach towards technologies? To see where and how the military element appears in the case of AI, the article aims to identify how military AI is framed in the EU. It empirically distinguishes three frames: military AI as a transformative technology (military capability building); military AI as increased insecurity (assessment of international dynamics as based on an arms race); and military AI as uncontrolled automation (calls to establish human control over automating military technology). The frames reveal differing views towards technology, the international environment, and the EU response to both. It argues that the notion of in-betweenness evolves not only in terms of blurring the lines between civil vs military use of the dual-use technology. It is also about EU self-positioning, which differs in the three frames (from military to normative power) by demonstrating manoeuvres between them depending on specific temporal and political circumstances.